Frustum culling
Discarding everything outside the camera's field of view before it is ever handed to the GPU.
The camera's visible volume is a truncated pyramid — a frustum — bounded by six planes: four sides plus the near and far clip planes. The test compares each object's bounding sphere or box against those planes, and if it lies wholly outside any one of them the object never needs processing. That is a handful of dot products per object, effectively free next to the cost of drawing it.
Practically every engine does this as a matter of course, often hierarchically: if a whole node in a spatial tree falls outside the frustum, everything beneath it disappears in one step. What the technique does not solve is hidden geometry inside the view — that is occlusion culling's job — nor the detail level of what genuinely is visible, which is what LOD handles.